Clinical epidemiology and risk prediction of unsuccessful treatment outcomes in pulmonary tuberculosis with diabetes mellitus

BACKGROUND: This study aims to characterize the epidemiological and clinical features of pulmonary tuberculosis with diabetes mellitus (PTB-DM), identify risk factors for unsuccessful treatment outcomes, and develop predictive models to aid in outcome assessment for these patients. METHODS: Clinical data from 3886 pulmonary tuberculosis (PTB) cases treated at Liuzhou People's Hospital Affiliated to Guangxi Medical University between July 2017 and December 2023 were analyzed. A case-control study was conducted to investigate the epidemiological and clinical profiles of PTB-DM. A retrospective cohort study and the LASSO-Logistic regression model were employed to identify independent risk factors for unsuccessful treatment outcomes, and a risk nomogram prediction model was developed to assess the predictive performance of the model. RESULTS: From 2017 to 2023, the rates of diabetes mellitus among pulmonary tuberculosis patients were 7.14%, 9.44%, 13.96%, 12.37%, 11.78%, 14.45%, and 19.63%, respectively. The proportion of pulmonary tuberculosis with diabetes mellitus showed a significant upward trend (P < 0.001). Various parameters including bacteriological positivity rate, sputum smear positivity rate at baseline, secondary pulmonary tuberculosis,and clinical symptoms like hemoptysis and chest pain, as well as imaging findings such as pulmonary cavity and pleural effusion, and unsuccessful treatment outcomes were all notably higher in PTB-DM patients compared to non PTB-DM patients (P < 0.05). Risk factors for unsuccessful outcomes in PTB-DM patients included being aged 65 or older (OR = 7.334, 95% CI: 2.654-23.136), hemoptysis (OR = 9.245, 95% CI: 3.411-27.464), positive sputum smear at baseline (OR = 3.774, 95% CI: 1.451-10.388), pulmonary cavity (OR = 3.850, 95% CI: 1.426-10.721), elevated CKMB levels (OR = 1.074, 95% CI: 1.014-1.136), elevated ESR (OR = 1.049, 95% CI: 1.025-1.077), and elevated HbAlc (OR = 1.352, 95% CI: 1.105-1.678), while higher ALB levels (OR = 0.874, 95% CI: 0.778-0.977) was a protective factor. A nomogram model incorporating these variables demonstrated high discriminative ability (AUC = 0.959, 95% CI: 0.931-0.987), with good calibration as assessed by the Hosmer-Lemeshow goodness-of-fit test (P = 0.474). CONCLUSIONS: Pulmonary tuberculosis patients with diabetes mellitus exhibit more severe clinical manifestations and a higher risk of unsuccessful treatment outcomes compared to those without diabetes. The nomogram model developed based on selected predictors demonstrates excellent predictive performance and can assist clinicians in early risk stratification and individualized management of PTB-DM patients.